View
1
Download
0
Category
Preview:
Citation preview
Quantification of Health and Productivity of Salt Quantification of Health and Productivity of Salt
Marshes Using Satellite DataMarshes Using Satellite Data
Hyun Jung (“J.”) ChoHyun Jung (“J.”) Cho
In collaboration with
D. Mishra, UGA
Professor,
Department of Integrated Environmental Science
Bethune-Cookman University
choh@cookman.edu
386 481 2793
Salt MarshSalt Marsh
The Deepwater Horizon oil spill
• 20 April 2010
– Continued flowing for
three months
• Largest off-shore oil
spill in U.S. historyspill in U.S. history
– Estimated 206 million
gallons of crude oil
• First major U.S. oil
spill to affect
marsh/wetland
habitathttp://gallery.usgs.gov/photos/05_05_2010_s84Aq10Ppk_05_05_2010_1
• Particularly damaging to marshland and
marine habitat
– Loss in green biomass (browning)
– Reduction in photosynthetic activity
Impact on Salt Marsh Habitat
– Impact from cleaning efforts
• Burning
• Flushing
• Skimming
• Estimate short term impact of the oil spill on salt
marsh habitat by evaluating marsh biophysical
characteristics
• Using remote sensing data, develop models and
Objectives
• Using remote sensing data, develop models and
products that will facilitate monitoring
restoration efforts of the coastal salt marsh
habitat
Study Area
Vegetation:Spartina patensSpartina alternifloraJuncus roemerianusDistichlis spicataSalicornia virginica
• 69 locations; Aerial and Ground Survey
• Top of Canopy (TOC) hyper-spectral reflectance data (Ocean optics USB
4000 Spectroradiometer and ASD sensor)
• Vegetation Fraction (Olympus E-502 Digital Camera)
Field DataAerial survey
• Vegetation Fraction (Olympus E-502 Digital Camera)
• Leaf Area Index (LAI Plant Canopy Analyzer 2000)
• Leaf level chlorophyll content (Minolta SPAD 502)
• Canopy level chlorophyll content calculating as Chlupper × LAI
• Above ground green biomass (gm/ft2)
Field MethodsDual Sensor Approach
↓
↑=E
L
ρ
L = upwelling radiance
Data acquisition
Sensor-Target : 16 ftIFOV: 2.2 meter
L = upwelling radiance
E = downwelling irradiance
p = remote sensing reflectance
(sr-1)
~2m
Field Methods:Canopy Chlorophyll
Leaf Level Chlorophyll
Leaf Area Index (LAI)
Canopy Chl = Leaf Level CHL * LAI
Leaf Area Index (Foliage area/Ground area)
The measurement of LAI is of fundamental importance in
ecological research because LAI is a measure of plant
growth; it directly affects the interception and absorption
of light by the canopy and it influences the primary
productivity of vegetation.80
LAI = 5.45
0
10
20
30
40
50
60
70
400 450 500 550 600 650 700 750 800 850 900
Wavelength (nm)
Refl
ec
tan
ce (
%)
LAI = 0
LAI = 5.45
Vegetation Fraction (%)
Vegetation Fraction (VF) = ratio of green vegetation area to
ground area
Vegetation fraction is obtained as a ratio of the number of
vegetation pixels to the total number of pixels in the image,
expressed in percentexpressed in percent
VF = 86% VF = 42%
1. Canopy Height
– Average of 5 measurements within a 1m radius of the scan center
2. Green Biomass
– Destructive sampling of vegetation after reflectance acquisition
• Samples were sorted, oven dried, and weighed
Sorting biomass
Biomass sample collection
Field DataN = 69 Min Max Mean St. Dev
Biomass
(g/ft2)3.9 566.2 82.6 84.04
LAI 0.01 2.57 1.02 0.7
VF (%) 0.3 99 34 25.34
40
50
60
70
80
90
100
Reflectance (%)
Calibration N = 69
VF (%) 0.3 99 34 25.34
Canopy
Chlorophyll
(mg/m2)
0.4 1321 134.1 209.88 0
10
20
30
400 500 600 700 800 900Reflectance (%)
Wavelength (nm)
Objective: Measure in-situ canopy-level reflectance of dominant vegetation
a) Develop spectral models to characterize selected biophysical parameters of each vegetation
species individually
b) Develop a single model to characterize biophysical parameters for the entire marsh
regardless of speciation
Applied Vegetation Indices (VIs) for GBM estimation
per species
Vegetation Index Application Formula Reference
Normalized Difference Vegetation Index (NDVI) Biomass, Chl, VF Rouse et al., 1974
Wide Dynamic Range Vegetation Index (WDRVI) Chl, LAI, VF Gitelson, 2004
Applied 15 VIs
- 5 models with 3 variants of each model using 760, 800, and 1100nm for NIR
Enhanced Vegetation Index (EVI) Biomass, LAI Huete, 2002
Chlorophyll Index – Green (CIgreen) Chl, GPP, VF Gitelson, 2006
Chlorophyll Index – Red Edge (CIred edge) Chl, GPP, VF Gitelson, 2006
Uses three characteristics of vegetation reflectance:
1) Visible wavelength with least absorption and most reflectance (rgreen)
- Does not saturate at high biophysical
Weighted Difference Biophysical Index (WDBI):
best correlated when working with all species
- Does not saturate at high biophysical values
2) Visible wavelength with the highest absorption (rred)
- Representative of non-pigment scattering
3) Spectral region most sensitive to vegetation canopy structure (rNIR)
Satellite Data
Multi-temporal LANDSAT TM 2009 - 2010
Dates 2009 2010
April 4 7
May 6 9May 6 9
June 7 3
July 9 5
August 3 6
September 4 7
October 13 9
Model Calibration
WDRVI= (α*λNIR – λRed) / (α*λNIR – λRed)
R2 = 0.53N = 69
R2 = 0.52N = 69
8
10
12
14
Perc
en
t N
orm
ali
zed
RM
SE
Model Validation
11.64 %
6.92 %
0
2
4
6
CHL GBM
Perc
en
t N
orm
ali
zed
RM
SE
6.92 %
Results
Results
Results
300
350
400
450
500D
ec
rea
se
in
are
a (
sq
.km
)2009
2010
0
50
100
150
200
250
CHL GBM
De
cre
as
e in
are
a (
sq
.km
)
Results: Phenology
Fringing Marshes Interior Marshes
0
10
20
30
40
50
60
70
80
90
April May June July August September October
Temperature (˚F)
2009
2010
Temperature
0
0.05
0.1
0.15
0.2
0.25
0.3
April May June July August September October
Pre
cip
ita
tio
n (
in)
2009
2010
Precipitation
Preliminary Results: MODIS
Preliminary Results: MODIS
• This study successfully delineate the critical hotspots and the pattern of marsh stress and so that prioritization of restoration areas can be performed
Conclusion and Future Research
• Tune models with more local data
• Apply the approach on the marsh degradation caused by other factors
This project was funded by National Science
Foundation RAPID program (Award # 1050500) and
Northern Gulf Institute-Mississippi State University
(Award # 306896)
Recommended